At the Honors College at USF, we work as web developers and IT support technicians, often staff members borrow equipment. Keeping track of over 250 items can be challenging and often expensive technology gets misplaced, lost, and even stolen sometimes.
1 in 11 people have shoplifted in the US---27million criminals *Roughly 1-1.2 million shoplifting Daily *Cost the public 33.21 billion yearly* *Current Solutions involve Security Cameras and Better Inventory Management. Is there a better Solution?
What it does
Track-A-BULL allows people to walk into an establishment, pick out objects and either borrow or purchase the item, depending on the establishment implemented in.
How we built it
We split into two teams. Two people worked on the Front End of the Web App and the Database, while the other members worked on the machine learning, dragonboard, and google cloud platform integration. Through this process, we were all exposed to new technology from NoSQL to Tensor Flow to using the Dragonboard with a camera.
Challenges we ran into
The Dragonboard was not setup ideally at first, which caused many connection errors. Also, some hardware lab components were originally missing from packages and led to some confusion.
Accomplishments that we're proud of
Tensor Flow- Machine Learning with OpenCV: Facial and Object Recognition
What we learned
Hackathons are fun, stressful, tiring, but extremely beneficial. Tensor Flow Firebase NoSQL
What's next for Track-A-Bull
Scalability, Authentication, More Precise Video Tracking, Training a model while capturing video, More Front End Features like search functionality